(Forthcoming in the Erkenntnis special issue on “Disagreements,” eds. T. Marques & D. Cohnitz)
DISAGREEMENT, RELATIVISM AND DOXASTIC REVISION
J. Adam Carter
ABSTRACT: I investigate the implication of the truth-relativist’s alleged ‘faultless disagreements1’ for issues in the epistemology of disagreement. A conclusion I draw is that the type of disagreement the truth-relativist claims (as a key advantage over the contextualist) to preserve fails in principle to be epistemically significant in the way we should expect disagreements to be in social-epistemic practice. In particular, the fact of faultless disagreement fails to ever play the epistemically significant role of making doxastic revision (at least sometimes) rationally required for either party in a (faultless) disagreement. That the truth-relativists’ disagreements over centred content fail to play this epistemically significant role that disagreements characteristically play in social epistemology should leave us sceptical that disagreement is what the truth-relativist has actually preserved.
1. Introduction: Two Debates About Disagreement Disagreement plays important philosophical roles in current debates in social epistemology and in the philosophy of language, though for very different reasons. In the philosophy of language, arguments from faultless disagreement are used to motivate truth-relativism over contextualism in certain areas of discourse. In social epistemology, disagreement is at the centre of debates between conformists and non-conformists about doxastic revision (in the face of disagreement with a recognised epistemic peer). What can these debates learn from each other? This is a largely unexplored question. My aim here is to develop a particular strand of connection between these debates; specifically, I explore the implications for the debate between conformism and non- conformism in social epistemology if the truth-relativist’s picture of disagreement is right. As we’ll see, the implications are more substantial than one might originally suspect. Toward the end of exploring these implications, I want to first (in §2) give a clear, general picture of the role disagreement plays in arguments for truth relativism, and in §3 I’ll outline and discuss just how disagreement over centred content is supposed to work, both in theory in practice. In §4, I’ll explain briefly what is at stake between conformists and non- conformists in the epistemology of disagreement, focusing on the role recognized peer
1 Faultless disagreement-style arguments are typically used motivate truth-relativism in various domains of discourse. For canonical presentations of faultless disagreement arguments for truth- relativism, see Kolbel (2003) and MacFarlane (2007). See also M. Kolbel and M. Garca-Carpintero (Eds.), Relative Truth, Oxford: Oxford University Press, for a collection of recent papers defending and opposing arguments from faultless disagreement. Cf. Cappelen & Hawthorne (2010) for a recent challenge to these arguments, and to truth-relativism more generally. 1 disagreement is supposed to play in doxastic revision. In §5, I’ll explore the question of whether centred-content disagreements appealed to by truth-relativists can play the same epistemically significant role with respect to questions about whether doxastic revision is rationally required in the face of such recognized disagreements with epistemic peers.
2. Truth-relativism, contextualism and disagreement Considerations to do with disagreement bolster an argument characteristic of a recent wave of what I’ll call “New Relativism2” in the philosophy of language. New Relativists endorse truth-relativism in certain areas of discourse where the truth of claims seems to depend, as MacFarlane puts it, ‘not only on how things are with the objects they explicitly concern, but on how things are with some subject not explicitly mentioned.’ Examples of such areas of discourse that have been fertile ground for New Relativists are: predicates of personal taste3, epistemic modals4, knowledge attributions5, future contingents6, gradable adjectives7 and indicative conditionals8. The most fertile ground for New Relativism has been discourse involving predicates of personal taste, and for ease of exposition, I’ll articulate the truth- relativist’s position vis-à-vis this particular area of discourse. Let P represent the area of discourse that is predicates of personal taste (such as tasty, fun). Truth-relativism with respect to P-utterances claims that (roughly): the truth of one’s P- utterance depends in part on a context of assessment—the context in which the proposition is being evaluated as true or false. Accordingly, “Miniature golf is fun,” does not get a truth- value simpliciter; it gets a truth value only once, as Lasersohn (2005) puts it, a judge or standards parameter (in the context of evaluation) is specified. And the relevant standard at play in the context of assessment is not uniquely determined by the context of use. As Crispin Wright (2007) has put it: vary [the context of assessment] and the truth value of the utterance can vary, even though the context of its making and the associated state of the world remain fixed. This is a somewhat simplistic view of the truth-relativist position, but it will do for now. Even the briefest reflection on the implications of truth-relativism reveals the position to constitute a radical departure from orthodoxy. Why exactly would one endorse truth-relativism in any domain of discourse? After all, as Cappelen & Hawthorne (2010)
2 I’m following here Maria Baghramian’s (forthcoming) terminology; Cappelen & Hawthorne (2010) refer to this recent movement in analytic philosophy of language “analytic relativism” while Wright (2007) refers to the position as “New Age Relativism.” 3 For some defences of truth-relativism about predicates of personal taste, see Lasersohn (2005), Stephenson (2007) and Kölbel (2003). Cf. Cappelen & Hawthorne (2010), Ch. 4, for some recent opposition. 4 See (Egan 2007; Egan, Hawthorne & Weatherson 2007; Stephenson 2007 and MacFarlane 2011c). 5 For defences of truth-relativism about knowledge attributions, see Richard (2004); MacFarlane (2005; 2010). 6 See MacFarlane (2003). 7 See Richard (2004). 8 For example, Weatherson (2006). 2 point out, the idea of the same proposition being true for Sam but not for Dean runs deeply contrary to our ordinary thinking about the contents of thought and talk—particularly, that the objects of thought and talk (i.e. propositions) bear truth and falsity as monadic properties. There would have to be very good reason to give up this elegant and pre-philosophically intuitive picture. MacFarlane thinks there are plenty of good reasons, and although he is not by any means the only New Relativist, his defences of truth-relativism across a spectrum of domains of discourse have been the most sophisticated in the literature; accordingly, I’ll be considering the role of disagreement in the truth-relativist’s argument broadly on MacFarlane’s own terms, and again, by focusing on predicates of personal taste (hereafter, P-utterances) as the representative area of discourse—though I will at times use examples from discourse concerning epistemic modals. As MacFarlane sees things, disagreement and subjectivity are twin desiderata that must be accommodated by any theory of the truth-conditions of P-utterances (and in other domains of discourse where truth-relativism is defended). To be explicit, a theory of the truth-conditions of P-utterances succeeds only if accounting for both
Subjectivity: how the truth of P-utterances depends in part on how things are with some subject not explicitly mentioned; and Disagreement: how when some subject A makes a P-utterance “ɸ” and B replies “Not ɸ”, this exchange constitutes a genuine disagreement.
Toward this end, the contextualist fares quite well in so far as the goal is to preserve subjectivity. For the contextualist, the proposition “Kalamata olives are tasty”, uttered by A, encodes A’s standards; accordingly, the proposition expressed by this utterance is (something like) “Kalamata olives are tasty (to A).” This already is a mark in favour of contextualism over invariantist approaches, which fail to account for how “Kalamata olives are tasty” is (when true) true in part due to facts about the utterer’s own tastes. The contextualist appears to be in trouble, though, when it comes to the second desiderata: preserving the insight that there can be genuine P-disagreements of the form:
A: ɸ (e.g. Miniature golf is fun) B: ~ ɸ (e.g. Miniature golf is not fun)
For the contextualist, A’s utterance expresses the proposition that miniature golf is fun to A, while B’s assertion expresses the proposition that miniature golf is not fun to B; accordingly, for the contextualist, there is no commonality of content about which A and B disagree. So the contextualist has no way to account for genuine P-disagreements. Or so the argument goes. As MacFarlane (and the contextualist) recognizes, the contextualist is not dead in the water at this point. Drawing from Lewis (1989), the contextualist could—following here 3
DeRose (2002)—invoke the shared scoreboard analogy in an effort to save disagreement9. On this analogy, the relevant standard in play is whatever standard is operative in the conversational context, and this standard can shift during the course of the conversation as different conversational moves are made. Consider here the case of knowledge-attributions: once (say) a particularly scary sceptical scenario is introduced, the standard operative in the conversational context raises, and claims of the form “S knows p” encode (after the scenario is introduced) that standard; because A and B can coherently assert and deny knowledge- attributions relative to the common standard operant in the conversational context (so long as they are keeping an eye to the scoreboard), we can make sense of A and B genuinely disagreeing. There are (at least) two problems with this initially promising approach for the contextualist to ‘regain’ disagreement. The first is that it is not at all clear how the shared scoreboard-strategy—while perhaps plausible in the case of knowledge-attributions and perhaps also in cases of epistemic modals—would work in P-discourse where it’s unclear what if any the standard operative in the apparent disagreement would be. Modifying here an example from Cappelen & Hawthorne, suppose Vinny (the vulture) says, “Rotting flesh is delicious!” while Gordon Ramsay replies that it is “(Expletive) terrible!” It is, at best, unclear how, for the contextualist, reference to the shared scoreboard analogy is supposed to save disagreement here. But even if the shared scoreboard strategy could be effectively employed here, a bigger worry looms: as MacFarlane notes, such a strategy could (in principle) only explain only intra-conversational (but not inter-conversational) disagreement. Here’s MacFarlane:
Once the importance of accounting for disagreement has been conceded, one cannot limit oneself to disagreement within conversations. And it is hopeless to widen the bounds of “conversations” as needed to make all disagreement intraconversational. For it is only if conversations are bounded and relatively self-contained that we can really make sense of the idea of a shared scoreboard. (MacFarlane 2007, 7.)
Just as there is no apparent ‘shared scoreboard’ in the cases where P-claims come from wildly different perspectives (as in the case of Vinny and Gordon Ramsay), there’s no shared scoreboard when disagreements are interconversational. What, after all, would a shared scoreboard amount to in inter-conversational disagreement? Here it is worth briefly considering ‘Eavesdropper’ cases: one subject overhears, and denies, a claim made in a conversational context of which she is not a recognised member. Eavesdropper cases have been especially forceful in suggesting that the contextualist cannot account for disagreement in cases where the truth of an epistemic modal is at issue (and this has been cited as an argument in favour of truth-relativist semantics for discourse about epistemic modality). A paradigmatic example of a claim expressing an epistemic modal is
9 Cf. Sundell (2011). Thanks to an anonymous referee for drawing my attention to this reference. 4 the claim A might be F; whether an utterance of this claim is true depends on whether F is an epistemic possibility for some relevant individual or group—that is, F must not be ruled out by what some individual or group knows10. As for which individual or group is relevant here, Egan (2011, 2), and Weatherson & Egan (2012, 4) point out that it depends, and (as the truth-relativist suggests, contra the contextualist) what it depends on does not seem to be uniquely determined by the context of utterance. Consider here a simple eavesdropping case, a variation on John Hawthorne’s (2007: 92) example:
EAVESDROPPER: Peeking out of my window, I overhear Rodney say to Jared, “Purdue might beat Kansas in the second round of the NCAA basketball Tournament!” Purdue was in fact leading throughout until Kansas made a very late game-winning comeback, which I know about, but they don’t (they quit watching while I remained in my flat vigorously clicking ‘refresh’ on ESPN). Even though it’s compatible with what they know that Purdue beat Kansas, I say to myself that what Rodney said was incorrect.
This looks like a legitimate (albeit inter-conversational) disagreement about whether Purdue might beat Kansas, but the contextualist seems to have no obvious way to account for it. If the relevant individual or group always (as DeRose suggests) includes the speaker11, then disagreement is lost because Rodney’s assertion expresses an epistemic possibility relative to a body of knowledge that is his (or perhaps his and Jared’s), and vice-versa for me. (After all, I am inclined to agree that Purdue’s winning is an epistemic possibility relative to what they know.) And, again, there is no shared scoreboard to make use of because this conversation is inter-conversational; thus, the contextualist fails to capture any clear sense in which Rodney and I disagree about whether Purdue might win. At this point, the truth-relativist turns a disagreement-based argument against contextualism into a disagreement-based argument for truth-relativism. While the invariantist is equipped to preserve disagreement without subjectivity, and the contextualist seems equipped to preserve subjectivity but not disagreement, the relativist claims you have it both ways: both subjectivity and disagreement. First subjectivity: as Stojanovic (2009, 691) succinctly notes, while the contextualist preserved subjectivity by modelling (for example, with predicates of personal taste) the judge parameter as an implicit argument to the taste predicate, the [truth]-relativist account models it as a parameter of the circumstances of evaluation.12
10 Or, perhaps, by whatever is within an agent’s epistemic reach. As Egan (2007, 11) suggests, “It might be the case that P is true iff it’s compatible with all of the facts that are within some group’s epistemic reach that P, where what it takes to be within one’s epistemic reach can vary across contexts. 11 See Egan, Hawthorne & Weatherson (2005) for further problems for DeRose’s speaker inclusion constraint. 12 Since truth-relativism is prominent in both predicates of personal taste and about epistemic modals, let X represent either area of discourse. As MacFarlane puts it, the subjectivity consists in the fact that 5
Whether “John might be at the store” is true, when evaluated by S, depends in part on what S’s knowledge rules out; and whether “Roller coasters are fun” is true, as evaluated by S, depends in part on what S judges to be fun13. It should be obvious now why, if truth- relativism can also really preserve disagreement (which we’ll consider shortly), then the disagreements that are preserved will be faultless: in the Eavesdropper case, while Rodney and I seem to disagree about whether “Purdue might win,” neither of us is at fault, or as Kölbel (2003; 2004) puts it, neither of us has “made a mistake.” The same would of course be the case with predicates of personal taste, and the fact that neither you nor I seem to be mistaken when asserting and denying (respectively) “Rollercoasters are fun” is an insight Kölbel thinks is both a ubiquitous one and one that only a truth-relativist can preserve. So subjectivity (and by extension faultlessness, in apparent disagreements) is preserved. But is genuine disagreement really preserved? This is a more complicated question. The first thing the relativist will draw attention to is the sense in which the truth- relativist preserves disagreement “better” than the contextualist. On the surface, this seems right. When Vinny and Gordon Ramsay disagree about whether rotting flesh is delicious, the contextualist, by viewing Vinny and Gordon as expressing propositions that encode their standards, respectively, is forced to say that there is no one proposition Vinny asserts and Gordon denies. Because the truth-relativist preserves subjectivity not by viewing ‘delicious’ as a hidden indexical, but rather, by including a standards parameter in the circumstances of evaluation, the truth-relativist allows Vinny and Gordon to affirm and deny the same standards-neutral proposition (i.e. Rotting flesh is delicious), respectively. Two things should now be obvious: firstly, that if we can’t make sense of genuine disagreement about the sort of content the truth-relativist claims is being ‘disagreed upon,’ then the motivation for endorsing truth-relativism over contextualism (across the domains of discourse where truth-relativism stakes its claim) is vitiated. Truth-relativism fails to account for disagreement if what it accounts for isn’t genuine disagreement. Secondly, it should be obvious that the interesting ramification for debates in social epistemology (specifically, in the epistemology of disagreement) will be whether doxastic revision in the face of truth-relativist-style disagreement with a recognised epistemic peer is something that we can make sense of as ever being rationally required. If not, truth-relativist disagreement would seem to lack the sort of epistemic significance disagreements as such are supposed to
the 'X-neutral' proposition (i.e. “John might be at the store” or “Roller coasters are fun”) which A asserts and B denies has a truth value only relative to an assessment parameter; as such, either of these propositions can be true for A and false for B. Subjectivity is preserved then because it is preserved that the truth of these claims depends in part on how things are for some agent not explicitly mentioned. 13 This is somewhat oversimplistic; as Lasersohn (2005) points out, in discourse about predicates of personal taste, we can distinguish between autocentric and exocentric perspectives, the latter of which will be ones where we the relevant standards at play in the context of evaluation will be the standards of someone other than the asserter. Consider, for example, a case where a man (having taken his 5-year-old to the carnival) says to his wife (in the presence of his son), that the “The rollercoaster was fun, but and other kiddie rides weren’t fun.” 6 have. We’ll engage with this latter issue after more carefully assessing the nature of truth- relativist disagreement.
3. Centred-content disagreement, and the practice of challenging assertions Lewis (1980) and Kaplan (1989) showed that truth can be relative to a context of evaluation that includes world, time and location parameters. As Kaplan (particularly in his work on indexicals) showed us, the need for particular parameters in the circumstance of evaluation is a function of the non-specificity of certain propositional contents with respect to world, time and location. Here’s Kaplan:
“A circumstance will usually include a possible state or history of the world, a time, and perhaps other features as well. The amount of information we require from a circumstance is linked to the degree of specificity of contents and thus to the kinds of operators in the language... (1989: 502, my italics)
MacFarlane, as well as Cappelen and Hawthorne14, view truth-relativism as expanding on this insight, by ‘proliferating’ the parameters in the circumstances of evaluation to include (as Cappelen & Hawthorne (2010, 10) put it), ‘exotic parameters’. Here’s MacFarlane’s own characterisation of this sort of parameter proliferation (which corresponds with a widening of the contents that are, for some parameter α, ‘α -neutral’:
Taking this line of thought a little farther, the relativist might envision contents that are “sense-of-humor neutral” or “standard-of-taste” neutral” or “epistemic-state” neutral,” and circumstances of evaluation that include parameters for a sense of humor, a standard of taste or an epistemic state. This move would open up room for the truth value of a proposition to vary with these “subjective” factors in much the same way that it varies with the world of evaluation. The very same proposition--say, that apples are delicious-- could be true with respect to one standard of taste, false with respect to another.” (MacFarlane 2007: 6-7)
What do two people disagree about when disagreeing about a (for example) ‘standard-of-taste-neutral’ proposition? Such a proposition is what some refer to a centred proposition, which—following Egan—is true only relative to a triple that includes 14 In their monograph-length argument against truth-relativism (which they call ‘analytic relativism’), Cappelen & Hawthorne write: “Contemporary Analytic relativists reason as follows: ‘Lewis and Kaplan have shown that we need to relativize truth to triples of relative to t, i>a differ only with respect to the parameter fixed by A’s and B’s standards (respectively). So, again, what do A and B disagree about when asserting and denying respectively a centred proposition? Isn’t the question ‘is [insert centred proposition] such and so?’ more sensible for either to ask (and assess) in soliloquy and less sensible for either to ‘debate’15? This is a tenuous predicament for the truth relativist, and MacFarlane knows it. 16 Accordingly, MacFarlane rejects the simple view of disagreement according to which A and B disagree if there is a proposition p that A accepts and B rejects. The simple view (vis-à-vis centred content) really doesn’t seem to capture genuine disagreement. Ragnar Francén, when critiquing Kölbel’s contention to the contrary, puts the worry well: Kölbel is right that on his view, when Kant says that punishing innocent Irwin is wrong and Smart says that this is not so, the propositions they assert cannot both be true at one circumstance of evaluation (moral perspective). But this does not mean that they disagree. In analogy with Jane’s and June’s assertions, Smart’s and Kant’s assertions concern different circumstances of evaluation, different moral perspectives. It might very well be that they agree that punishing innocent Irwin is wrong at Kant’s moral perspective and that punishing innocent Irwin is not wrong at Smart’s moral perspective. (Francén 2008, 112, cited also in Dreier 2009, 99). MacFarlane himself, in rejecting that two people disagree just so long as there is a proposition that one accepts and the other rejects, offers a more sophisticated accuracy- based account of disagreement, supplemented by a formal account of accuracy: Disagreement (MacFarlane): Two parties disagree (as assessed from context C) if: (i) There is a proposition that one party accepts and the other rejects, and (ii) The acceptance and the rejection cannot both be accurate (as assessed from C). I will spare the details here of MacFarlane’s account of perspectival accuracy on which this account of disagreement is premised17. This is because, given what MacFarlane ultimately concedes about his account, we can set aside the matter of whether the account is 15 Perhaps as sensible as it is to debate what fixes the judge parameter when it’s you that is evaluating the proposition; and so, perhaps as sensible as debating what your own standards are. 16 Cf. Egan 2011 for an assertion-based attempt, on behalf of the truth-relativist, to take this problem on. 17 MacFarlane views Perspectival Accuracy as a component of his account of disagreement, which he calls Can’t Both Be Right. According to Perspectival Accuracy: An acceptance (rejection) of a proposition p at a context Cu is accurate (as assessed from a Context CA) iff p is true (false) at the circumstance Wcu, SCA where Wcu = the world of Cu and SCA = the standard of taste of the assessor at CA. 8 ‘counterexample-free.’ My aim at this point18 isn’t to assess whether truth relativism gives a correct account of disagreement, but rather, to consider what implications such an account would have for debates about disagreement in social epistemology. And here, what is more important than his formal account of accuracy is the relationship he thinks accuracy (in his account) bears to the practice of challenging assertions that is part and parcel with disagreement in practice. On this point, recent work by James Dreier (2009, 102) reveals some very awkward results for the truth-relativist. Following Dreier, we can characterise how challenging assertions is supposed to work on a relativized account of accuracy. Accuracy/Challenge: (i) one is entitled to challenge an assertion when one has good grounds for thinking that the assertion was not accurate (relative to the context of assessment one occupies in issuing the challenge), and (ii) a successful response to such a challenge consists in a demonstration that the assertion was, in fact, accurate (relative to the context of assessment one occupies in giving a response’) Against this background, Dreier (2009, 102) assesses how this Brandom-style challenge/response game, which MacFarlane (2007, 29) admits ‘can look like a pretty silly game’, would work vis-à-vis relativized accuracy: Well, we could play this game. I shout, “It is throbby in here,” and you challenge. You assess “It is throbby in here” from your own perspective and, pain free, find it badly defective (inaccurate); you find you were entitled to your challenge. I now roll my eyes, place my head in my hands, grope around for Ibuprofen, thus demonstrating that from my perspective my assertion was entirely accurate; I have successfully (from my perspective) met the challenge. We could play this game. But it would be a bad game. Dreier (2009, 102-3) goes on to suggest that not only is this a bad game, but even worse, perhaps a nonsensical game. The game as we were playing it seems not to make any kind of conversational sense at all. In this way it is quite different from other kinds of pointless conversational games we might play… [other pointless games] are conversationally intelligible even if their critics are right about their pointlessness. When you say that human beings are four- dimensional and I say we are three dimensional, this at least makes sense as a disagreement even if there isn’t anything either of us can say to budge the other form his position… but when I insist that it is throbby in here and you reply that it isn’t the least 18 In §5, I’ll give some considerations to suppose not that the truth-relativist account of disagreement is ‘incorrect’, but rather, that it is not epistemically significant in the way we should expect genuine disagreements to be. 9 bit throbby, all that is happening is that I have a headache, and you do not, and we both know this… and there is no intelligible sense in which we disagree. Perhaps Dreier is right that these faultless ‘disagreements’ are nonsensical, perhaps he’s not. They do seem to be practically pointless, and for good reason: the alleged disagreement here lacks any epistemically significance for either party. Genuine disagreements are hardly pointless in practice in part because they serve the function of playing epistemically significant roles in social-epistemic practice, as we’ll see in §4. In §5, we’ll consider whether truth-relativist disagreement can play the sort of epistemically significant role genuine disagreements are taken to play. 4. Disagreement in social epistemology: conformism versus non-conformism Is the mere fact that you and I disagree about p epistemically significant in the following way: if I previously believed p, and then find out that you believe ~p, I am rationally required to revise (i.e lower the degree of my belief19) my belief that p? A quick and easy answer is ‘no’, if I (for example) know that my evidence is better than yours, or perhaps, I know that you are delusional, or in some way cognitively defective on the matter at hand. Such asymmetry with respect to our epistemic standings to p prior to disagreement is enough to make our disagreement epistemically insignificant with respect to whether, rationally, I am required to revise my belief that p. The fact of our disagreement (in such circumstances) doesn’t provide me with any rational grounds for supposing that I am wrong about p. But what if, prior to disagreeing about whether p, we recognise each other as cognitive equals, or epistemic peers with respect to whether p? What then? This depends in part on what epistemic peerhood involves. Feldman and Lackey suggest that we are epistemic peers (vis-à-vis p) if we are on a par with respect to our cognitive ability and evidence relevant to determining whether p20. If I recognise you as an epistemic peer prior to a disagreement with you about p, recognition of this fact of disagreement is perhaps not insignificant with respect to what rationality requires of my doxastic attitude with respect to p. Adam Elga, a proponent (along with Feldman and David Christensen) of what Lackey calls the conformist view21, says: 19 Or perhaps withhold judgment with respect to p. 20 Lackey calls these criteria for epistemic peerhood evidential equality and cognitive equality. On her articulation: Evidential equality: A and B are evidential equals relative to the question whether p when A and B are equally familiar with the evidence and arguments that bear on the question whether p. Cognitive equality: A and B are cognitive equals relative to the question whether p when A and B are equally competent, intelligent, and fair-minded in their abilities to assess the evidence and arguments that bear on the question whether p. Cf. Earl Conee (2009) “Peerage,” Episteme 6 (3):313-323 for a thorough discussion of epistemic peerhood. 21 Christensen (2007) calls this the ‘conciliatory’ view. 10 On my usage, you count your friend as an epistemic peer with respect to an about to-be- judged claim if and only if you think that, conditional on the two of you disagreeing about the claim, the two of you are equally likely to be mistaken. Elga (2007, 487) And once this is supposed, as Elga (2007, 478) sees it, the rational thing to do is straightforward: you should give [the recognised epistemic peer’s] conclusions the same weight as your own. Here’s Elga: Suppose that before evaluating a claim, you think that you and your friend are equally likely to evaluate it correctly. When you find out that your friend disagrees with your verdict, how likely should you think it that you are correct? The equal weight view says: 50%. Elga’s is a strong view of conformism. More moderate versions of conformism will require some doxastic revision in the face of disagreement with a recognised epistemic peer. David Christensen (2007, 4) argues that the fact of disagreement with a (recognised) epistemic peer rationally requires me to ‘change my degree of confidence significantly’ toward that of my peer, and vice-versa, while, according to Richard Feldman, what rationality requires (in the face of recognized epistemic peer disagreement) is that I suspend judgment. Feldman and Elga (and to a large extent Christensen), thus, take something like an ‘all or nothing’ view: what rationality requires of me, upon finding out that an epistemic peer believes contrary to me about whether p, is doxastic revision to the extent that I no longer maintain positive belief that p. If this is right, an important implication of the conformist position is the denial of the claim that there can be any reasonable disagreement among mutually recognised epistemic peers (at least in situations of full disclosure). This picture is rejected by what Jennifer Lackey (2007, 4-7) calls non-conformists (e.g. Kelly (2005); van Inwagen (1996)). Non-conformists, specifically, reject the view that one is rationally required to revise one’s belief in the face of recognised peer disagreement, and consequently, unlike the conformist, allow for the possibility of rational or reasonable peer disagreement. Who’s right, the conformist or the non-conformist? As literature over the past five years has shown22, there is no quick and easy answer. Offering one is more than I want to attempt here. However, I do want to draw attention, as Lackey has, to the fact that certain cases are typically adduced in favour of the conformist and certain others in favour of the non-conformist. But this very fact, I think, supports Lackey’s contention that we have some reason to reject the following sort of principle, which she calls uniformity: 22 See Disagreement, ed. Feldman & Warfield, (2010) for an anthology of important recent work on the epistemology disagreement. Also, cf. Lackey “The Epistemology of Disagreement,” in Oxford Bibliographies Online. 11 Uniformity: Disagreement with epistemic peers functions the same epistemically in all circumstances. Both conformism and non-conformism presuppose uniformity. But if uniformity is false, then there is scope for intermediate positions (such as what Lackey endorses herself) according to which we allow that, in some cases, the rationally appropriate thing to do in the face of recognised peer disagreement is to revise one’s belief; in others, it is rationally appropriate to remain steadfast23. The challenge for developing an intermediate position is to spell out in some principled way what grounds different rational requirements in response to recognised peer disagreement in different cases. I’m interested here not in evaluating Lackey’s preferred intermediate account24, but rather, to suggest that if some category of cases seems to predominantly support conformism, then this is a prima facie reason to think that rationality requires at least some degree of doxastic revision in this class of cases, even if there is some other class of cases that typically squares better with non- conformism. A rather wide class of cases that has been popular among conformists is the (broad) class of cases that include disagreements between recognised epistemic peers concerning controversial subject matter.25 Let’s say a subject matter S is controversial to the extent that epistemic peers tend to disagree in that area26. Such areas include exclude (for instance) basic arithmetic, the current time, physical laws, geography, and to a large extent empirical science and include (among other areas) politics, religion, philosophy and—most relevant for our purposes—claims of taste. It should be uncontroversial that claims characteristic of these areas are controversial, and apparent disagreement about them is rampant. One might be tempted to reflect, though, on this last suggestion and point out: “While it’s fair enough to imagine A and B as 23 An example of a case that would plausibly support a non-conformist approach would be one where the previously recognized epistemic peer disagrees with you about something for which your epistemic justification is overwhelming, direct and immediate. For example, suppose 24 See Lackey’s “A Justificationist Account of the Epistemic Significance of Peer Disagreement.” 25 Thanks to an anonymous referee for pointing out that controversial issues have formed some of the core examples for all participants in the literature on peer disagreement, whatever their preferred view. Perhaps this is, to some degree, a result of non-conformists attempting to meet the challenge posed by conformists, when the challenge is presented vis-à-vis controversial subject matter (a la Feldman). However, as the reviewer points out, this isn’t always the case. Whether and to what extent controversial subject matter is comparatively more fertile ground for the conformist than for the non- conformist is an interesting philosophical question in its own right. For the present purposes, I do not need to weigh in here; my position depends only on the weaker claim that controversial subject matter is typically used to support the non-conformist verdict, and this is so regardless of whether non-conformists appeal in some cases to similar kinds of cases. 26 Another perhaps equally plausible characterization is: a subject matter is controversial to the extent that the relevant experts in that subject matter disagree about propositions central to that subject matter. Plausibly, in practice, these characterisations will pick out the same subject matter. Even if not, either sufficiently picks out claims of taste as controversial subject matter. 12 epistemic peers engaged in a disagreement about politics or philosophy, it’s not clear how A and B could be epistemic peers with respect to some taste-claim. People can disagree about taste claims, but never as epistemic peers.” In reply, we may point out: for anything two people can disagree about, they can disagree about as epistemic peers and potentially recognise each other as such (and so the doxastic revision issues between conformists and non-conformists arise). After all, if A and B disagree about whether P is true, then (prior to the disagreement), along dimensions of intellectual ability vis-à-vis determining whether p and evidence vis-à-vis p, one of the three situations will be the case: (i) A is in a better epistemic position along these dimensions for determining whether p, and so A is B’s epistemic superior; (ii) B is in a better epistemic position along these dimensions for determining whether p than A, and so A is B’s epistemic inferior; (iii) A is in a relevantly similar epistemic position (to B) along these dimensions for determining whether p, and so A and B are epistemic peers. If A and B are epistemic peers about whether p, then given suitable disclosure of their positions along the ability and evidential dimensions, A and B can recognise each other as epistemic peers about whether p, prior to disagreeing about p. Thus, if A and B can disagree about p, one way they can disagree about p is as recognised epistemic peers, and then the question arises: is doxastic revision rationally required in the face of this disagreement? And since claims of taste fall squarely in a category of the sort of cases that typically motivate conformism27: a plausible answer is “yes.” If we can make sense of how rationally required revision of beliefs would work in the face of peer disagreements about taste, when the disagreement works as the truth-relativist sees it, then great (for the truth- relativist’s picture of disagreement)! If not, though, the faultless-disagreement-style cases to which truth-relativists appeal fail to have the sort of epistemic significance disagreements (qua disagreements) are supposed to have in social epistemology. Plausibly, an account of disagreement must be amenable to accounts of the characteristic roles disagreements play in practice, including their epistemic roles. To the extent that faultless disagreement (as championed by the truth- relativist) fails to play such a role, there is reason to cast doubt on the matter of whether these are disagreements at all. And if they are not, then all the worse for the truth-relativist, who—as we saw in §2—loses traction against the contextualist to the extent that 27 See fn. 25. Also, thanks to an anonymous referee for pointing out that, in aesthetic disputes particularly, a non-conformist position could be potentially be motivated in part by considering that disputants behave in practice as though non-conformism were correct. Does the fact that disputants act as though non-conformism is true (by in fact holding their guns) in area X favour non-conformism in area X as a position about the rationality of doxastic revision in area X? It’s not clear that it does any more than it suggests, as Feldman would contend, that most disputants hold their guns unreasonably in these areas. 13 disagreement is lost. Taking a step back, it should be clear now why the relationship between disagreement in the philosophy of language and social epistemology is actually something on which a good deal hangs. That said, let’s consider one more avenue of resistance to the inter-debate gambit I’ve sketched for truth-relativist disagreement: suppose one denies that (in the arena of claims of taste) there could be recognised peer disagreement because there is no proper disagreement at all in this area because, for example, disagreement over centred content isn’t genuine disagreement, or, following the expressivist, such claims express attitudes and are not truth- apt in a way that allows for genuine disagreement (thus nor recognised peer disagreement)). To this line, we may reply: perhaps that’s right. Maybe there is no genuine disagreement at all when A says “This is tasty” and B says “This is not tasty.28” But no position has been taken here. To be clear: what I’m interested in is how to make sense of the social-epistemic side of disagreement if we take the truth-relativist at her word. The truth relativist tells us (emphatically, in debates with the contextualist) that these are disagreements. And as I’ve suggested, for any p, if two people can disagree about whether p, they can do so as recognised epistemic peers. And if they do so as epistemic peers, then if the debate is in an area apposite to conformism, truth-relativist-style disagreement had better be such that we can make sense of being rationally required to revise our beliefs in light of recognised (truth- relativist) peer disagreement. Let’s turn now to the question of whether such sense can be made. 5. Truth-relativist disagreement and doxastic revision Let F be the claim “Bob Saget is funny,” sincerely expressed by Libby. Libby says “F,” and Libby’s respected colleague, Eunice, who typically shares Libby’s sense of humour, replies with ~F. As the truth-relativist tells us, “F”—that is, “Bob Saget is funny”—is true for Libby and false for Eunice. They also tell us it is a disagreement. If it is a disagreement, then it should play the relevant epistemic role that disagreements play in social epistemology. Since any disagreement about whether p can in principle be a disagreement individuals approach as epistemic peers vis-à-vis p, and since this particular subject matter is one that plausibly squares with conformism as an approach to epistemic peer-disagreement, the truth-relativist owes us an explanation for the following, supposing Libby and Eunice are recognised epistemic peers vis-à-vis F: (R) Rationality requires that, if Libby and Eunice disagree about whether F and are mutually recognised epistemic peers vis-à-vis F, then in light of the fact that they disagree vis-à-vis F, each should revise her respective belief about whether F. 28 Dreier’s observations (considered in §3) about the apparent futility of the practice of challenging (and successfully replying to) assertions, within the frame-work of truth-relativism could be taken to support such a suggestion. 14 As we noted in §4 in considering Elga’s presentation of the motivations for conformism, a principle like (R) is premised upon a related principle about recognised epistemic peerhood, call it (EP): (EP) If A and B are recognised epistemic peers vis-à-vis F, then rationality requires that A and B think, before disagreeing about F, that each other is equally likely to be right about whether F. If the truth relativist is to make sense of (R), the truth relativist must be able to make sense of (EP). Can the truth-relativist make sense of (EP)? It’s clear to see that when the disagreement is about the truth of a traditional proposition, (EP) would seem right: If A and B recognise each other as epistemic peers, then they would be rationality required to think (prior to disagreement) each other is equally likely to be right about F (otherwise, it’s not clear that they have recognised each other as epistemic peers). It’s much less clear though how we can make much sense of (EP) in the context of the truth relativist’s faultless disagreement. Take again Libby and Eunice and the centred (taste- neutral) proposition “Bob Saget is funny,” which gets a truth value only relative to a ‘judge’ parameter—Libby, when Libby is evaluating the proposition, Eunice, when Eunice is evaluating the proposition. Let’s now suppose the antecedent of (EP) is satisfied, and A and B are recognised epistemic peers vis-à-vis F. This means that Libby and Eunice recognise each other as satisfying Lackey’s cognitive equality and evidential equality criteria for epistemic peerhood: so each recognises the other as equally intelligent and capable on the matter at hand—they both recognise each other as having watched all the same Saget programs and having (say) read all the same literature on humour. Does it follow that Libby and Eunice are rationally required to accept each other as equally likely to be right about whether F? Importantly, no. For the truth relativist, the connection between epistemic peerhood and likelihood of getting it right breaks down. This is because cognitive equality and evidential equality don’t entail standards equality, and standards equality is just as relevant to getting it right, for the truth-relativist. (After all, the sort of claims being disagreed about are true relative not only to the state of the world but also to the evaluator’s standard). So the truth- relativist can’t coherently endorse (EP), but without endorsing (EP), it’s unclear how the truth-relativist is supposed to make sense of (R). An uncomfortable upshot, then, is that, two individuals can recognise each other as epistemic peers vis-à-vis p by recognising each other to be cognitively and evidentially symmetrical vis-à-vis p and yet (i) not be in a position to rationally conclude (prior to the disagreement) that each other is equally likely to be right about p; and (ii) never be rationally required by the fact of this disagreement to revise their beliefs about whether p. 15 A general result, then, is that: the truth relativist cannot explain how epistemic peer disagreement is ever epistemically significant; for persons A, B, if A and B recognise each other as epistemic peers vis-à-vis p and then find out that they come to different conclusions about whether p, this fact will never make it rationally required for A to revise her belief about whether p. Given that both conformism and Lackey’s intermediate position on peer disagreement support a conformist reading (on which at least some doxastic revision is rationally required) in cases where we have recognised peer disagreements about taste- claims, the fact that the truth-relativist’s faultless disagreements lack the ability to play this plausible social-epistemic role disagreements play (in such cases) qua disagreements is a strike against faultless disagreements; these alleged disagreements are not epistemically significant in the way disagreements are thought to be in social epistemology. But given the way the challenge/assertion game would be played, this is just what we’d expect. The fact of disagreement itself would lack any epistemic significance; as we saw Dreier pointed out, the parties in the dispute, in response to the mutually recognised fact of disagreement (peer or otherwise), would have no reason to do anything other than steadfastly repeat themselves in a way that constitutes a successful meeting of the challenge, again and again (ad nauseum). The situation gets a bit worse, though. Recall that whenever A and B genuinely disagree about whether p, then (prior to the disagreement), along Lackey’s cognitive and evidential dimensions relevant to correctly evaluating whether p, A will be either B’s epistemic peer, inferior or superior. So if taste disagreements are disagreements about whether something is so, they can be disagreements approached in such a way that one party is a recognised epistemic superior of another: that is, if what’s at issue is whether p, one party can be better cognitively and evidentially positioned to determining whether p. Suppose A and B stand in a starkly asymmetrical relationship along both of these dimensions vis-à-vis p. Suppose further A and B, over a cup of tea, are keenly aware of the asymmetry. A says p, but B says ~p. B here is A’s (recognised) epistemic superior vis-à-vis whether p. Is doxastic revision rationally required for A in the face of disagreement with a recognised epistemic superior? It seems so. As Bryan Frances, says: It may seem obvious that upon such a discovery I should at the very least reduce my confidence level in P by some significant amount, if not withhold belief entirely or move closer to her view than my old view. (Frances 2012, 2) Whether doxastic revision is rationally required in the face of disagreement with a recognised epistemic peer has generated much more discussion than the same question regarding epistemic superiors, because—with epistemic superiors—the question seems to be ‘how much’ doxastic revision is rationally required, rather than ‘whether’ it is required. If things were otherwise, we would have a hard time explaining why it is irrational (all things 16 equal) to be doxastically unmoved by the testimony of experts29. In the face of disagreement with a recognised epistemic superior, it is prima facie plausible to suppose you are rationally required not only to revise your belief, or to withhold, but to positively move your credence towards the view of the expert. Because this last bit is somewhat controversial 30, let’s suppose something considerably weaker, and so: (R*) Rationality requires that, if A and B disagree about whether p and B is A’s recognised epistemic superior vis-à-vis p, then in light of the fact that they disagree vis-à-vis p, A should (to some extent) revise her respective belief about whether p. And (R*) is premised upon something like: (EP*) If B is A’s recognised epistemic superior vis-à-vis p, then rationality requires that A think, before disagreeing about p, that A is (at least somewhat) more likely to be right about p than B. It would be quite bad if the truth-relativist’s faultless disagreements were not compatible with (R*). But truth-relativist disagreement is compatible with (R*) only if compatible with EP*. But truth-relativists disagreements are not even in principle compatible with EP*. This is because cognitive superiority (vis-à-vis p) and evidential superiority (vis-à-vis p) don’t entail standards superiority (or even standards equality) vis-à- vis p, and standards (which fix the judge parameter) are just as relevant to getting it right, for the truth-relativist. One might balk and suppose that that no two individuals can enter into a taste agreement where one is an epistemic superior, and so the worry that truth-relativists can’t preserve EP* (and R*) regarding taste disagreements is circumvented. But this worry can be countered the same way we countered the same worry mutatis mutandis for epistemic peers. The reply is that, if A and B genuinely disagree about whether something is so, then they may or may not be symmetrical with respect to their cognitive abilities and evidence that are relevant to determining whether p. If a faultless disagreement is a disagreement, then qua disagreement, it should (in a context where there is stark asymmetry between A and B along cognitive and evidential dimensions vis-à-vis p) be epistemically significant. But such disagreements never will be. 6. Conclusion 29 Consider here Elga’s (2007, 479) discussion of epistemic superiority in a weather forecasting case: “How, exactly, should we be guided by outside opinions?... Start with the simplest case: complete deference. When it comes to the weather, I completely defer to the opinions of my local weather forecaster .My probability for rain, given that her probability for rain is 60%, is also 60%.” 30 Cf. Frances 17 Although disagreement plays (very) different roles in debates in the philosophy of language and in social epistemology, these debates could stand to learn from each other, and I hope to have drawn attention to one important point of connection. In particular, I’ve suggested that any attempt to ‘save’ disagreement by the truth-relativists in the philosophy of language should aspire to save something capable of playing the characteristically epistemically significant role disagreement is supposed to play in social-epistemic practice; I’ve given reason to doubt that the sort of disagreement the truth-relativist appeals to as an advantage over the contextualist could ever play such an epistemically significant role. I leave open whether the worry I raise is sufficient for rejecting truth-relativism, all- things-considered; I insist only that we should be sceptical of any argument for truth- relativism that relies on premises that suggest the truth-relativist has preserved disagreement. From the perspective of social epistemology, it seems the epistemic significance of disagreement is preserved by the truth-relativist no better than by the contextualist, and this should be concerning for the truth-relativist. Plausibly, the issue examined here is but one of many points of connection between disagreement in social epistemology and the philosophy of language, and it would benefit both debates if more points of connection were explored. REFERENCES Baghramian, M. (forthcoming) “The Many Faces of Relativism,” Il relativismo, Temi e prospettive, Vittorio Villa (ed), forthcoming. Brandom, R. (1983). “Asserting,” in Nous, Vol. 17, No. 4 (Nov., 1983), pp. 637-650. Cappelen, H. & Hawthorne, J. (2009) Relativism and Monadic Truth (Oxford: Oxford University Press). Carter, J. A. (2012) “A Note on Assertion, Relativism and Future Contingents,” in Logos & Episteme, 3: 1 (2012). Christensen, D. (2007). “Epistemology of Disagreement: the Good News,” The Philosophical Review 116: 187-217. Cohnitz, D. Conceptual and Faultless Disagreements (draft) Conee, E. (2009) “Peerage,” Episteme 6 (3):313-323 DeRose, K. (2004). “Single Scoreboard Semantics,” in Philosophical Studies, 119, 1-21. 18 Dreier, J. (2009). “Relativism (And Expressivism) and the Problem of Disagreement,” in Philosophical Perspectives, 23, Ethics, 2009. Egan, A., Hawthorne, J. & Weatherson, B. (2005). “Epistemic Modals in Context,” In Gerhard Preyer and Georg Peter (eds.) Contextualism in Philosophy: Knowledge, Meaning and Truth, 131-170. Oxford University Press. Egan, A. (2007) “Epistemic Modals, Relativism and Assertion,” Philosphical Studies 133: 1-22. —— (2011) “Relativism about Epistemic Modals” in A Companion to Relativism: Blackwell Companion to Philosophy, ed. Stephen Hales: Wiley-Blackwell Publishing. Egan, A. & Weatherson, B. (2011) Epistemic Modality, eds. Egan & Weatherson, Oxford: Oxford University Press. Elga, Adam (2007) “Reflection and Disagreement,” Nous 41: 478-502. Evans, G. (1985): ‘Does Tense Logic Rest on a Mistake?’ in Collected Papers, pp. 343-363, Oxford: OUP. Feldman, R. (2007) “Reasonable Religious Disagreement,” in Louise Antony (ed.), Philosophers without God: Meditations on Atheism and the Secular Life, Oxford: Oxford University Press. —— “Epistemological Puzzles about Disagreement,” Epistemology Futures, edited by Stephen Hetherington, Oxford, 2006, pp. 216-36. Francén, R. (2010). No Deep Disagreement for New Relativists. Philosophical Studies 151:19--37 Frances, B. (2010). The Reflective Episternic Renegade. Philosophy and Phenomenological Research 81 (2):419-463 —— (2012). “Discovering Disagreeing Epistemic Peers and Superiors,” International Journal of Philosophical Studies 20 (1):1 - 21. Hawthorne, J. (2007) “Eavesdroppers and Epistemic Modals” in Philosophical Issues 17, The Metaphysics of Epistemology, 2007. Kaplan, D. (1989). “Demonstratives: An Essay on the Semantics, Logic, Metaphysics, and Epistemology of Demonstratives and other Indexicals,” in Themes from Kaplan, ed. J. Almog, J. Perry, and H. Wettstein (Oxford: Oxford University Press), 481-566. Kelly, T. (2005) “The Epistemic Significance of Disagreement,” in John Hawthorne and Tamar Szabo Gendler (eds.), Oxford Studies in Epistemology, Vol. 1, Oxford: Oxford University Press Lewis. D (1979) “Scorekeeping in a Language Game,” Journal of Philosohpical Logic, 8, 339-359. 19 —— (1980) Index, Context, and Content. In Stig Kanger & Sven Öhman (eds.), Philosophy and Grammar, Reidel. Kölbel, M. (2002) Truth Without Objectivity; London: Routledge. Kölbel, M. (2003) “Faultless Disagreement” Proceedings of the Aristotelian Society New Series, Vol. 104: 53-73 Lasersohn, P. (2005) “Context Dependence, Disagreement, and Predicates of Personal Taste,” Linguistics and Philosophy, 28: 643-86 MacFarlane, J. (2007). “Relativism and Disagreement,” Philosophical Studies 132: 17-31. —— (2005) “The Assessment Sensitivity of Knowledge Attributions,” in T. Szabo Gendler and J. Hawthorne (eds.) Oxford Studies in Epistemology 1: 197-233. —— (2011a) “Simplicity Made Difficult,” Philosophical Studies 156 (3):441-448. —— (2011b) “Epistemic Modals are Assessment Sensitive,” in Epistemic Modality, eds. Egan & Weatherson, Oxford: Oxford University Press. —— (2011c) “Epistemic Modals are Context Sensitive,” in Epistemic Modality, ed. Egan, A. & Weatherson, B, Oxford: Oxford University Press. —— (2008) “Truth in the Garden of Forking Paths,” In M. Kolbel and M. Garca-Carpintero (Eds.), Relative Truth, pp. 81{102. Oxford University Press. —— (2010) “Relativism and Knowledge Attributions”, in Routledge Companion to Epistemology, ed. Sven Bernecker and Duncan Pritchard (London: Routledge, 2010) Stephenson, T. (2007). “Judge Dependence, Epistemic Modals, and Predicates of Personal Taste,” Linguistics and Philosophy, 30: 487-525. Stojanovic, I. (2007). “Talking about Taste: Disagreement, Implicit Arguments, and Relative Truth,” Linguistics and Philosophy 30 (2007): 691-706. Sundell, T. (2011). “Disagreements about Taste,” Philosophical Studies 155 (2011): 267-288. Weatherson, B. (2011) “No Royal Road to Relativism,” Analysis (2011) 71 (1): 133-143. —— (2009) “Conditionals and Indexical Relativism,” Synthese 166: 333-57. —— (2001) “Indicative and Subjunctive Conditionals,” Philosophical Quarterly 51: 200-16. Wright, C. (2007). New Age Relativism and Epistemic Possibility: The Question of Evidence, Philosophical Issues 17 (1):262--283. 20